linux-kernel - Re: [PATCH v13 13/18] x86/tsc: calibrate tsc only once

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CAGM2reb-MwiT8qoiSkXCjAAM46JOYL9BP+U3q7PMTcxp9EiK7w@mail.gmail.com>
Date:   Fri, 13 Jul 2018 07:30:39 -0400
From:   Pavel Tatashin <pasha.tatashin@...cle.com>
To:     douly.fnst@...fujitsu.com
Cc:     Steven Sistare <steven.sistare@...cle.com>,
        Daniel Jordan <daniel.m.jordan@...cle.com>,
        linux@...linux.org.uk, schwidefsky@...ibm.com,
        Heiko Carstens <heiko.carstens@...ibm.com>,
        John Stultz <john.stultz@...aro.org>, sboyd@...eaurora.org,
        x86@...nel.org, LKML <linux-kernel@...r.kernel.org>,
        mingo@...hat.com, tglx@...utronix.de, hpa@...or.com,
        peterz@...radead.org, prarit@...hat.com, feng.tang@...el.com,
        Petr Mladek <pmladek@...e.com>, gnomes@...rguk.ukuu.org.uk,
        linux-s390@...r.kernel.org, boris.ostrovsky@...cle.com,
        jgross@...e.com
Subject: Re: [PATCH v13 13/18] x86/tsc: calibrate tsc only once

On Fri, Jul 13, 2018 at 3:24 AM Dou Liyang <douly.fnst@...fujitsu.com> wrote:
>
>
> At 07/12/2018 08:04 AM, Pavel Tatashin wrote:
> > During boot tsc is calibrated twice: once in tsc_early_delay_calibrate(),
> > and the second time in tsc_init().
> >
> > Rename tsc_early_delay_calibrate() to tsc_early_init(), and rework it so
> > the calibration is done only early, and make tsc_init() to use the values
> > already determined in tsc_early_init().
> >
> > Sometimes it is not possible to determine tsc early, as the subsystem that
> > is required is not yet initialized, in such case try again later in
> > tsc_init().
> >
> > Suggested-by: Thomas Gleixner <tglx@...utronix.de>
> > Signed-off-by: Pavel Tatashin <pasha.tatashin@...cle.com>
>
> Hi Pavel,
>
> Aha, a complex solution for a simple problem! ;-) And I did find any
> benefits of doing that. did I miss something?

Hi Dou,

I had this in previous version:  init early, and unconditionally
re-init later (which required to resync sched clocks for continuity,
and check for some other corner cases). Thomas did not like the idea,
as it is less deterministic: it leads for code to work by accident,
where we might get one tsc frequency early and another later, and so
on. The request was to initialize only once, and if that fails do it
again later. This way, if early initialization is broken, we will know
and fix it.

>
> As the cpu_khz and tsc_khz are global variables and the tsc_khz may
> be reset to cpu_khz. How about the following patch.

Could you please explain where you think this patch can be applied,
and what it fixes?

Thank you,
Pavel

>
> Thanks,
>         dou
> ------------------------8<-----------------------------------
>
> diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c
> index 74392d9d51e0..e54fa1037d45 100644
> --- a/arch/x86/kernel/tsc.c
> +++ b/arch/x86/kernel/tsc.c
> @@ -1370,8 +1370,10 @@ void __init tsc_init(void)
>                  return;
>          }
>
> -       cpu_khz = x86_platform.calibrate_cpu();
> -       tsc_khz = x86_platform.calibrate_tsc();
> +       if (!tsc_khz) {
> +               cpu_khz = x86_platform.calibrate_cpu();
> +               tsc_khz = x86_platform.calibrate_tsc();
> +       }
>
>          /*
>           * Trust non-zero tsc_khz as authorative,